Article https://doi.org/10.

1038/s41467-023-39236-4

Quality evaluation of ground improvement

by deep cement mixing piles via ground-
penetrating radar
Received: 5 February 2023 Hongyan Shen 1,2 , Xinsheng Li 3, Ruifeng Duan 3,4 , Yong Zhao5,
Jing Zhao1,2, Han Che1,2, Guoxin Liu1,2, Zhijia Xue6, Changgen Yan6, Jiwei Liu6,
Accepted: 2 June 2023
Chao Jiang7, Boke Li3, Hong Chang3, Jianqiang Gao6 & Yueying Yan1

Check for updates Deep cement mixing piles are a key technology for treating settlement distress
1234567890():,;
1234567890():,;

of soft soil subgrade. However, it is very challenging to accurately evaluate the

quality of pile construction due to the limitations of pile material, large
number of piles and small pile spacing. Here, we propose the idea of trans-
forming defect detection of piles into quality evaluation of ground improve-
ment. Geological models of pile group reinforced subgrade are constructed
and their ground-penetrating radar response characteristics are revealed. We
have also developed ground-penetrating radar attribute analysis technology
and established ground-penetrating radar technical system for evaluating the
quality of ground improvement. We further prove that the ground-penetrating
radar results integrating single-channel waveform, multi-channel section and
attributes can effectively detect the defects and stratum structure after
ground improvement. Our research results provide a rapid, efficient and
economic technical solution for the quality evaluation of ground improvement
in soft soil subgrade reinforcement engineering.

Subgrade settlement is a serious distress problem faced by the
construction of highways and railways1–7. It is particularly promi-
nent in soft loess covered areas, river valley sections, mined out
areas, karst development areas, etc., which poses a huge threat to
the service life and operation safety of roads. Therefore, in the
process of subgrade construction, it is necessary to take effective
measures by every means to improve the quality of subgrade con-
struction to prevent the occurrence of subgrade settlement
distress8–11. Deep cement mixing (DCM) pile is a kind of cement plus
solid, which uses cement as the main curing agent and forcibly
mixes the soft soil and cement paste in the deep part of the foun-
dation through mixing machinery, making the soft soil harden into
a cement reinforced soil with integrity and certain strength,
thereby improving the strength of the foundations12–16. It has the
characteristics of low specific gravity of consolidated body, large
bearing capacity, low permeability coefficient, and has the advan-
tages of low cost, short construction time and maximum utilization
of the original soil. Therefore, it has been widely used in soft soil
ground improvement, foundation settlement treatment and other
engineering construction around the world17–24. At present, DCM
piles have become one of the key technologies in highway distress
prevention and treatment engineering. However, some piles may
have different degrees of defects due to the dynamic conditions of
soil layer and groundwater, construction technical defects, con-
struction loopholes and other reasons. It has laid hidden dangers
for inducing accidents25,26. Therefore, pile defect detection and pile

1
School of Earth Sciences and Engineering, Xi′an Shiyou University, Xi’an 710065, P.R. China. 2Shaanxi Key Laboratory of Petroleum Accumulation Geology,
Xi’an 710065, P.R. China. 3Shaanxi Geo-mining Geophysical and Geochemical Exploration Team Co. Ltd., Xi’an 710043, P.R. China. 4National Engineering
Research Center of Offshore Oil and Gas Exploration, Beijing 100028, P.R. China. 5Shaanxi Land Construction Surveying, Planning and Design Institute Co.
Ltd., Shaanxi Land Construction Group, Xi’an 710075, P.R. China. 6School of Highway, Chang′an University, Xi’an 710064, P.R. China. 7Gansu Luqiao Highway
Investment Co., Ltd., Lanzhou 730030, P.R. China. e-mail: shenhongyan@xsyu.edu.cn; 450991137@qq.com; Yanyueying@xsyu.edu.cn

Nature Communications | (2023)14:3448 1

Article
construction quality evaluation are important guarantee technol-
ogies for subgrade distress treatment.
The traditional engineering pile foundation detection mainly
adopts static load testing, drilling coring, high-strain and low-strain
dynamic measurement. Static load testing27–29 and drilling coring30,31
are relatively accurate and fair methods for pile foundation detection.
However, both of these methods are destructive testing techniques
with long cycles, high costs, low sampling ratios (1~2%) and poor
representativeness of sampling. High-strain dynamic measurement32–34
is to impact the pile top with a heavy hammer, and then analyze and
determine the vertical compressive bearing capacity and pile integrity
of a single pile based on the wave theory. This method can only provide
reference bearing capacity. Low-strain dynamic measurement35,36
involves applying a dynamic force (dynamic load) to the pile top, and
the structural integrity of the pile body is judged through observation
and analysis of dynamic response signals. This method has the
advantages of fast detection speed, low cost and good reliability, and
has a good detection effect for high-strength pile foundations, such as
cast-in-place piles, prefabricated piles, etc. Currently, pile foundation
testing is mainly based on this technology37–39. However, DCM piles are
locally sourced and mixed with cement paste to form piles. Although
there are differences between the elasticity of the piles and the original
stratum, the difference is far less obvious than that of the piles cast
with sand, cement and coarse aggregates as the main material13,29.
Therefore, the low-strain dynamic measurement technology devel-
oped on the basis of elastic theory may not be suitable for the detec-
tion of DCM piles. In addition, the number of DCM pile groups is too
huge, the pile spacing is small, and the pile group density is high for the
soft soil reinforcement engineering. It is difficult to obtain obvious
detection effect and high economic benefits by using this traditional
pile foundation testing technology.
Ground-penetrating radar (GPR) is a survey technology based on
electromagnetic field theory. Its basic principle is to transmit ultra-
high frequency (106~109 Hz) pulse electromagnetic waves, and then
analyze and infer the structure and physical properties of under-
ground media according to the changes of echo amplitude, waveform,
frequency and other characteristics by using the differences of elec-
tromagnetic properties of underground media40–43. It has the advan-
tages of high resolution, intuitive results and fast scanning speed. In
recent years, it has become an important technology and means for
various engineering quality detection, near- surface stratum structure
and defect survey44–55. Recent research has shown that there are sig-
nificant differences in electromagnetic properties between DCM pile
and original stratum, and there are also significant differences in
electromagnetic properties between defective and complete piles56,57.
Therefore, GPR technology is fully applicable to the quality inspection
of DCM piles.
Based on the above reasons, we proposed to use GPR technology
to detect the defects of DCM piles, and then realize the quality eva-
luation of ground improvement. Our innovations are to put forward
the idea of transforming defect detection of pile foundations into
quality evaluation of ground improvement, construct geological
models of subgrade reinforced by pile groups and reveal theirs GPR
response characteristics, develop a GPR multi-attribute information
extraction technology, and establish a GPR technical system for eval-
uating the quality of ground improvement. The pile length, strong/
weak reinforcement stratum, primary stratum, the buried depth and
range of defective formation can be effectively identified by integrat-
ing the GPR information of single-channel waveform, multi-channel
section and attributes. Our research results provide a fast, efficient and
economic way for the quality evaluation of DCM pile reinforcement for
soft soil foundation, and fill the technical gap of efficient detection of
pile foundation defects in the case of large number of pile groups with
small pile spacing and high pile group density. Moreover, our tech-
nology can also be expanded for pile foundation defect detection and
Nature Communications | (2023)14:3448
https://doi.org/10.1038/s41467-023-39236-4
pile construction quality evaluation in other projects under high-
density pile groups, such as slope protection, dam seepage prevention,
foundation reinforcement, and other engineering.
Results
Geological models and their GPR response characteristics
The purpose of using DCM piles in soft soil foundation is to enheance
the bearing capacity of the stratum and then to improve the stability of
the foundation. If the subgrade strengthened by DCM pile groups is
regarded as a complete set of strata, the problem of pile foundation
detection can be transformed into the problem of ground improve-
ment quality evaluation (Fig. 1a, b). If there are quality problems in
the foundation evaluation, it will also reflect that there are defects in
the pile foundations. The purpose of this is to create conditions for the
selection of continuous detection technologies. It can significantly
reduce the cost of pile foundation detection while effectively
improving the efficiency and effect of pile foundation testing.
In order to effectively use GPR to detect the quality of pile foun-
dations or evaluate the effectiveness of ground improvement, it is
necessary to first understand the propagation laws of electromagnetic
waves under different geological conditions of subgrade and the GPR
response characteristics. For this reason, we analyzed the electro-
magnetic waves propagation and GPR response characteristics of the
geological models of reinforced subgrade by pile groups. In the
foundations without pile (Fig. 1c), electromagnetic waves are reflected
only at the wave impedance interfaces (stratum interfaces). For the
foundations reinforced by complete piles (Fig. 1d), in addition to
forming reflections at the interfaces of wave impedances, electro-
magnetic waves also generate complex reflection, transmission,
refraction and guided waves between piles as well as between piles and
the original stratum. Thus, GPR response characteristics with relatively
stable frequency and amplitude are formed in the equivalent forma-
tion. For the foundations reinforced by defective piles (Fig. 1e), due to
defects in the piles or stratum (such as pile breaking, necking, segre-
gation, etc.), the energy distribution of electromagnetic waves at the
defect locations will be changed, which will also lead to change in GPR
response characteristics. Specifically, the amplitude become weaker
(type I response characteristics in Fig. 1e) or stronger (type II response
characteristics in Fig. 1e), and the frequency also become lower or
higher.
In the quality evaluation process of ground improvement,
although the subgrade strengthened by piles is equivalent to a set of
strata, the GPR response characteristics of equivalent strata are com-
pletely different from those of strata without pile, and there may be
significant differences between the GPR response characteristics of
defective piles and complete piles (Fig. 1c, d, e). Therefore, these GPR
response characteristics and differences become an important basis
for identifying whether piles have defects and evaluating the quality of
ground improvement.
Technical system
The outstanding characteristics of ground improvement by DCM piles
are that the project progress is fast, the construction process is closely
connected, and the time available for pile foundation defect detection
and ground improvement quality evaluation is limited. It is required
that pile foundation defect detection and ground improvement quality
evaluation not only to quickly and accurately detect the distribution
areas and scopes of defective piles or unfavorable stratum, but also to
meet the needs of rapid construction of highway engineering, so as
to timely feedback on the detection results and take timely measures
to supplement piles. Therefore, we have established a technical system
for using GPR to detect defects in DCM piles and evaluate the quality of
ground improvement based on engineering geological survey data and
highway construction site conditions (Fig. 2). Specifically, it includes
the following four aspects.
2
Article
Geological basis. Comprehensively master the stratum structure and
physical properties of subgrade as well as the thickness of soft soil
layer based on engineering geological survey data. Deeply understand
design the parameters such as pile material, diameter, length and
spacing as well as the technical scheme for subgrade construction.
Data acquisition. Design GPR survey lines or networks based on the
site conditions of highway construction. The basic principle is to lay
survey lines along the extension direction of the highway and along the
axis of the pile foundations (Fig. 1b). For pile foundation detection with
a length of several meters to more than ten meters, it is recommended
to select 100~200 MHz antennas and use profiling method to con-
tinuously collect GPR data. Note that the sampling length must com-
pletely include the equivalent layer depth of the reinforced subgrade
by pile foundations, and the sampling rate is generally from 0.1 to
0.4 ns to meet the detection requirements.
Data processing. Carry out fine processing of GPR data around key
links such as energy enhancement in deep layers, interference noise
suppression and multiple attenuation. Ensure that GPR data proces-
sing results with high signal-to-noise ratio, high fidelity and high
resolution are obtained.
Data interpretation. Integrate single-channel waveform, multi-channel
section and attribute information, and combine drilling results to
interpret geological information in GPR data. Realize the detection of
pile foundation defects and accurate quality evaluation of ground
improvement. It should be emphasized here that 1~2 boreholes can be
properly arranged on the GPR survey line (or network), rather than
many boreholes. The purpose of borehole coring is not only to verify
the GPR interpretation results, but also to calibrate the GPR section
with borehole results to further calculate and obtain more accurate
a b
GPR GPR
c d
1 2 3 1
Fig. 1 | Geological models of ground improvement by pile groups and their GPR
response characteristics. a Geological model of subgrade reinforced by pile
groups. b Equivalent geological model of subgrade reinforced by pile groups. Pile
groups are equivalent to a stratum, and defect detection of pile foundations is
Nature Communications | (2023)14:3448
https://doi.org/10.1038/s41467-023-39236-4
electromagnetic wave velocities to achieve time-depth conversion
processing of the GPR section.
Project overview and GPR data acquisition
At present, DCM piles have been widely used in the treatment of
subgrade settlement distress caused by soft loess settlement24,58–60. We
have carried out the application and practice of GPR technology in
defect detection of DCM piles and quality evaluation of ground
improvement based on the construction project of Anlin highway in
Gansu Province, China. Anlin highway stats from Anjiazui Village,
Taishi Town, Linzhao County, and ends at Baichuan Village, Dongyuan
Township, Linxia County (Fig. 3a), with a total length of 56.7 km. It
would be constructed according to the technical standards of two-way
four lane second-class highway.
It was found that there were signs of soft soil foundations in
Sanjiaji section during the highway survey stage. A combination of
drilling and excavation exploration was used to obtain foundation soil
samples. It was found by observing the foundation soil samples that
the undisturbed soil particles in this section were mainly composed of
silt particles (particle size 0.006~0.078 mm), with obvious porosity.
Macropores and wormholes were visible to the naked eye, and con-
tained a large mount of salt crystals. Therefore, it was determined that
this section belongs to typical soft loess. DCM pile technology was
selected to strengthen and treated the subgrade of this section in
order to ensure high-quality construction of highway engineering. The
design scheme for subgrade reinforcement is shown in Supplementary
Fig. S1. The DCM piles were arranged in a regular quincunx pattern,
with a pile diameter of 0.5 m, a pile spacing of 1.5 m, and a pile length of
10.0 m. A total of more than 10,000 DCM piles were built in the entire
highway section.
28 days after the completion of pile construction, the work of
using GPR technology to evaluate the quality of pile foundations and
Transmitting antenna
Receiving antenna
Ray path of direct waves
Ray paths of reflection at formation interfaces
Ray paths of electromagnetic waves about pile 1
Ray paths of electromagnetic waves about pile 2
Ray paths of electromagnetic waves about pile 3
Subgrade defect
GPR records
Pile
GPR I II
e
2 3
transformed into quality evaluation of ground improvement. c Stratum without
pile, (d) stratum with complete piles, (e) stratum with defective piles and theirs GPR
response characteristics. Type I denotes energy weakening, and type II represents
energy enhancement.
3
Article https://doi.org/10.1038/s41467-023-39236-4

Fully understand subgrade structure, physical properties and thickness of soft soil layer as well
Geological Basis as engineering design parameters, such as pile foundation material, diameter, length and
spacing of piles.

Design GPR survey line or network according to the situation of the project site. Select
100~200 MHz antennas, and use the profiling method to collect GPR data. Pay attention to the
Data Acquisition
selection of data acquisition parameters, such as GPR record length, sampling rate and point
spacing.

Carry out fine processing of GPR data. Pay attention to gain recovery, energy compensation,
Data Processing
interference noise suppression, multiple attenuation and topographic correction.

Analysis and interpretation of single-channel

Interpretation of GPR waveform characteristics
waveform and section Analysis and interpretation of multi-channel
section characteristics

Waveform attributes
Data Interpretation of GPR Amplitude attributes
Interpretation attributes Frequency attributes
Phase attributes

Verification of GPR interpretation results

Drilling and coring Velocity calculation
Horizon tracking

Defect detection of piles foundations and quality evaluation of ground improvement

Fig. 2 | Technical system for DCM pile foundation defect detection and ground improvement quality evaluation via GPR. The technical system includes four parts
that are geological basis, data acquisition, data processing, and data interpretation.

a b c

China

Lanzhou

Yongjing

Dongxiang Anjiazui
Baichuan
Sanjiaji
Linxia
Hezhen Linzhao
Kangle

Fig. 3 | Location of work area and field data acquisition. a Location of work area. b DCM pile construction scene. c GPR data acquisition scenario.

the effect of subgrade reinforcement was carried out. We used the SIR-
4000® GPR system, selected 100 MHz antennas, and used the profiling
method for GPR data acquisition. The scene of piling and field GPR
data acquisition is shown in Fig. 3b, c. The data acquisition parameters
included that the sampling rate was 0.351 ns, the sampling length was
1024 points per channel, and the channel spacing was approxi-
mately 2.0 cm.
Interpretation of GPR section
Taking a survey line in the exploration area as an example, the appli-
cation effect of our proposed technology is introduced in detail.
Supplementary Fig. S2 shows the original GPR data obtained for this
survey line and Fig. 4a shows the GPR data processing results
according to the data processing flow in Supplementary Fig. S3. Four
typical channels were extracted from the GPR section in Fig. 4a for

Nature Communications | (2023)14:3448 4

Article https://doi.org/10.1038/s41467-023-39236-4

Distance (m) A preliminary interpretation of the GPR section can be further

a 0 10 20 30 40 50
0 0 made based on the above single-channel waveform interpretation
results, and combined with the variation characteristics of waveform,
amplitude, frequency and phase of the section. The interpretation
100 5 results are shown in Fig. 4c, in which the area surrounded by a green

Depth (m)
Time (ns)

irregular circle was interpreted as a subgrade defect, the cyan solid line
was interpreted as a clear bottom interface of subgrade reinforcement,
200 10 and the cyan dotted lines were interpreted uncertain bottom inter-
faces of subgrade reinforcement.

300 15 Interpretation of GPR attributes

Distance (m) Although the quality of DCM piles and the effect of subgrade rein-
b 0 10 20 30 40 50
0 0 forcement can be preliminarily evaluated based on the single-channel
Completer pile waveform and multi-channel section characteristics of GPR data, some
Defect
Defect interpretation conclusions are ambiguous because the information
100 5 carried by GPR data has not been fully explored. As shown in Fig. 4c, it

Depth (m)
Time (ns)

Pile?
Pile? Pile bottom is difficult to give an accurate conclusion as to whether there are pile
Pile bottom? Pile bottom Pile bottom? foundations or defects in the left A and right B areas of the section.
200 10 Therefore, we used the GPR attribute information extraction and
analysis technology that we recently developed to further process the
C1 C2 C3 C4 section in Fig. 4a. It was attempted to clearly reveal whether the pile
300 15
foundation was complete and accurately evaluated the quality of weak
Distance (m)
c 0 10 20 30 40 50 subgrade reinforcement through GPR attribute information. Here, we
0 0
only selected one of the most representative GPR attributes from
waveform, frequency, amplitude and phase attributes categories for
in-depth analysis and geological information interpretation.
100 5
Depth (m)
Time (ns)

B
A Arc length of time window (AL). There must be differences in GPR
200
300
Bottom interface of piles (Determined) Defect boundary
Bottom interface of piles (Uncertain) Borehole
A, B Areas with unclear geological interpretations C1~C4 Labels of GPR channels
Fig. 4 | Interpretation of GPR section. a GPR data processing results. b Waveform
interpretation of typical channels. c Multi-channel section interpretation. Inter-
pretation of GPR section is performed after analyzing and interpreting the wave-
form characteristics of typical channels. It is not possible to determine whether pile
foundations exist in A and B areas. Source data are provided as a Source Data file.
waveform feature analysis and interpretation in order to accurately
grasp the GPR response characteristics. The extracted GPR channels
were identified as C1, C2, C3 and C4, respectively (Fig. 4b). In the GPR
channel marked with C1, the energy between 44 and 57 ns was weak
and the waveform changed irregularly. It was interpreted that there
were subgrade (or pile foundation) defects between 44 and 57 ns of the
channel based on the analysis conclusion in Fig. 1e. The energy
between 122 ns and 163 ns of the channel was also weak, but a wavelet
with strong energy appeared near 163 ns. Therefore, it was impossible
to judge whether the pile bottom was at 122 ns or 163 ns based solely
on the waveform characteristics. In the GPR channel labeled C2, there
was waveform disorder between 48 and 119 ns. This indicated that
there were defects in the subgrade (or pile foundation). The GPR
energy suddenly weakened at 163 ns of this channel, which should be
the GPR response characteristic of the pile bottom. In the GPR channel
labeled C3, the waveform changed were more regularly. It was judged
that the pile foundation was complete or there was no defect in the
subgrade based on the analysis conclusion in Fig. 1d. The amplitude
attenuation of the channel appeared at 131 ns, indicating that this
might be the pile bottom. In the GPR channel marked with C4, there
was a weak and chaotic energy phenomenon between 97 and 159 ns, so
it was impossible to determine whether the pile bottom was at 97 ns
or 159 ns.
response characteristics between reinforced and primary stratum,
10
between strong and weak reinforced stratum as well as between
defective and non defective stratum. However, when the differences in
15 GPR response characteristics are not obvious, it is difficult to distin-
guish them in conventional GPR sections, such as the left A and right B
areas in Fig. 4c. It could be seen from the AL attribute in Fig. 5a that the
boundary between strong and weak values clearly showed the bottom
interface of the strongly reinforced stratum (cyan solid line). The
interface was interpreted as a pile bottom interface in the GPR section
in Fig. 4c, which might be a wrong interpretation results. In addition,
the energy disturbance areas of this attribute section clearly deli-
neated the defective areas of the subgrade (irregular green solid line
circles), which was clearer and more accurate than that in the GPR
section in Fig. 4a.
Product of instantaneous amplitude and cosine of instantaneous
phase (PIACIP). It could see from the PIACIP attribute in Fig. 5b that
the effect was equivalent to that of the AL attribute in Fig. 5a. The
bottom interface of the strongly reinforced stratum was clear (cyan
solid line) as well as the defect areas of the subgrade were also clear
(irregular green solid line circles). It further confirmed that the inter-
pretation of the AL attribute was reliable.
Slope of instantaneous frequency (SIF). As could be seen from Fig. 5c
that the bottom interface of the strongly reinforced layer was not
reflected in the section of SIF attribute, but it clearly revealed the
interface between the weakly reinforced layer and the primary stratum
(bottom interface of the weakly strengthened layer, purple solid line).
The bottom interface information of the weakly reinforced layer could
not be obtained from the above two attributes in Fig. 5a, b and the
conventional GPR section in Fig. 4a. The disordered areas of this
attribute also clearly circled the defective areas of the subgrade (irre-
gular green solid line circles).
Slope of reflection strength (SRS). We could see from Fig. 5d that the
SRS attribute also accurately identified the boundary between the
weakly reinforced layer and the primary stratum (purple solid line),

Nature Communications | (2023)14:3448 5

Article https://doi.org/10.1038/s41467-023-39236-4

Distance (m) Distance (m)

a 0 10 20 30 40 50 b 0 10 20 30 40 50
0 0 0 0

100 5 100 5

Depth (m)
Depth (m)
Time (ns)

Time (ns)
200 10 200 10

300 15 300 15
Distance (m) Distance (m)
c 0 10 20 30 40 50 d 0 10 20 30 40 50
0 0 0 0

100 5 100 5

Depth (m)

Depth (m)
Time (ns)

Time (ns)
200 10 200 10

300 15 300 15
Bottom interface of strongly reinforced stratum Bottom interface of weakly reinforced stratum Defect boundary Borehole

Fig. 5 | Interpretation of GPR attributes. a Arc length (AL) attribute with an strongly reinforced layer is interpreted through AL and PIACIP attributes. The
analysis time window of 1.0 ns. b Product of instantaneous amplitude and cosine of bottom interface of the weakly reinforced layer is determined through SIF and SRS
instantaneous phase (PIACIP) attribute. c Slope of instantaneous frequency (SIF) attributes, and the subgrade defect areas are delineated by combining four types of
attribute. d Slope of reflection strength (SRS) attribute. The bottom interface of the GPR attributes. Source data are provided as a Source Data file.

with an effect equivalent to the SIF amplitude in Fig. 5c. The disordered
areas of the attribute also clearly delineated the defective areas of the
stratum (green irregular circles).
Comparison between GPR results and borehole coring
Based on the above interpretation of GPR data, we selected two
boreholes arranged near C2 and C4 in Fig. 4b for coring, which were
marked as Borehole #1 (Fig. 6a, b) and Borehole #2 (Fig. 6c, d),
respectively. And the depth of Borehole #1 and Borehole #2 were
approximately 11.0 m and 10.0 m, respectively.
The coring results for Borehole #1 were divided into five segments
(Fig. 6a, b). The first segment (0~2.37 m) and the third segment
(5.92~8.25 m) corresponded to the 0~47.3 ns and 119.5~163.5 ns seg-
ments in the GPR attribute sections in Fig. 5, respectively. It was shown
as gray cement block with good cementation and complete pile body.
The second segment (2.37~5.92 m) corresponded to the 47.3~119.5 ns
segment in the GPR attribute sections in Fig. 5. It was manifested as
broken cement blocks with large pores and poor integrity of the pile
body. This might be caused by the excessive acceleration of the nozzle
during construction, resulting in uneven mixing of cement and soil.
The fourth segment (8.25~9.69 m) corresponded to the 163.5~193.7 ns
segment in the GPR attribute sections in Fig. 5. It appeared as gray
cement blocks. The cementation property was not as good as that of
the first segment. And the integrity of the pile body was slightly poor.
The reason might be that the water content in the deep soil layer,
resulting in poor pile formation effect of DCM piles. The fifth segment
(9.69~11.0 m) was primary loess.
The coring results of Borehole #2 were divided into three segments
(Fig. 6c, d). The first segment (0~4.91 m) corresponded to the 0~97.2 ns
segment in the GPR attribute sections in Fig. 5, and the core sample
characteristics were the same as those of the first and third segments of
Borehole #1. The second segment (4.91~8.26 m) corresponded to the
97.2~164.5 ns segment in the GPR attribute sections in Fig. 5, and the
core sample characteristics were the same as the fourth segment of
Borehole #1. The third section (8.26~10.0 m) was primary loess.
It can be seen from the comparison between the above borehole
coring results and the GPR attribute interpretation results that the two
are basically consistent. Based on the drilling results and GPR inter-
pretation results (Table 1), we further calculated the electromagnetic
wave velocities. For the subgrade reinforced by DCM piles, there might
be some differences in the physical properties of the medium due to
the different construction quality at different locations, which might
lead to some deviations in the electromagnetic wave velocities
between them. However, this difference was relatively weak according
to the statistical results of Borehole #1 and Borehole #2 (Table 1).
Therefore, we took the average value of velocities (vave ≈ 0.1 m · ns−1) as
the final electromagnetic wave velocity. And all GPR data processing
results in Figs. 4, 5 were subjected to time-depth conversion proces-
sing based on this average velocity.
Evaluation of ground improvement
Based on the interpretation results of typical channels, multi-channel
section and attributes of GPR data (Fig. 6e) as well as constrained with
boreholes (Fig. 6a, b, c, d), we had constructed a geological model
(Fig. 6f) of the subgrade reinforced by DCM piles below this survey
line. The stratum after ground improvement could be divided into
three layers that were strongly reinforced layer, weakly reinforced
layer and primary stratum. The average buried depth of the bottom
interface of the strongly reinforced layer was approximately 6.0 m.
The shallowest distributed on the right side of the section, approxi-
mately 4.4 m. And the deepest was distributed in the middle left of the
section, approximately 8.5 m. The average buried depth of the bottom
interface of the weakly reinforced layer was approximately 9.0 m. The
shallowest was approximately 7.2 m, distributed in the middle right of
the section. And the deepest was approximately 10.0 m, distributed in
the middle left of the section. There were two defect areas in the
stratum of the reinforced subgrade, which were identified as D1 and D2,
respectively. The defect area D1 presented an irregular shape and was
distributed on the right side of the section. It had a large area, spanning
of approximately 24.0 m and a buried depth of 2.1~6.6 m. The defect

Nature Communications | (2023)14:3448 6

Article https://doi.org/10.1038/s41467-023-39236-4

Distance (m)
a c e 0 10 20 30 40 50
Seriously defective 0 0

D2
D1
100 5

Depth (m)
Time (ns)
A B

200 10
Borehole #1 Borehole #2
b Borehole #1 d Borehole #2
0.00 0.00 300 15
Distance (m)
f 0 10 20 30 40 50
2.37 0

Depth (m) D2
D1
4.91 5
Depth (m)

Depth (m)
5.92

8.25 8.26 10

9.69 10.00
11.00 15
Determined pile bottom interface (by GPR waveforms) Uncertain pile bottom interface (by GPR waveforms)
Bottom interface of strongly reinforced stratum (by GPR waveforms) Defect boundary (by GPR waveforms)
Bottom interface of weakly reinforced stratum (by GPR attributes) Defect boundary (by GPR attributes)
Complete Slightly defective Seriously defective Primary loess Borehole
Strongly reinforced stratum Weakly reinforced stratum Defective area Primary stratum
A, B Areas with unclear geological interpretations (by GPR waveforms) D1, D2 Areas with geological defects

Fig. 6 | Construction of geological model after ground improvement. the GPR interpretation results constrained by dilled results. The reinforced sub-
a, b, c, d Drilling and coring results of Borehole #1 and Borehole #2. e Comparison grade is divided into three layers that are strongly reinforced stratum, weakly
of interpretation results of GPR waveforms (section) and attributes. f Geological reinforced stratum and primary stratum. There are two defects in the subgrade that
model after ground improvement. The geological model is constructed based on are D1 and D2. Source data are provided as a Source Data file.

Table 1 | Calculating electromagnetic wave propagation velocities by combining GPR interpretation and drilling results
Stratum interface Borehole #1 Borehole #2
Depth /m Double-way travel Velocity /m · ns−1 Depth /m Double-way travel Velocity /m · ns−1
time /ns time /ns
Top interface of D1 2.37 47.3 0.1002 - - -
Bottom interface of D1 5.92 119.5 0.0991 - - -
Bottom interface of strongly reinforced 8.25 163.5 0.1009 4.91 97.2 0.1010
stratum
Bottom interface of weakly reinforced 9.69 193.7 0.1001 8.26 164.5 0.1004
stratum
Average velocity /m · ns−1 - 0.1 - 0.1

area D2 presented an approximately flat elliptical shape and dis-
tributed in the right side of the section. It spanned approximately
13.0 m, had a thickness of approximately 1.0 m and a buried depth of
2.2~3.3 m. Therefore, it was recommended to supplement piles to the
defect D1 and D2 areas.
Discussion
DCM piles have played a key role in the treatment of foundation set-
tlement distress17,19,21–24. However, piling is a complex and systematic
construction process. Small errors in each construction step may
cause deviations between the pile body and the design, such as
insufficient (or uneven) mixing of cement paste and deep soil particles
as well as the lack of coordination between the lifting speed of the
grouting pipe and the grouting amount, etc. Even if the pile
construction process is carried out in accordance with standard pro-
cess as well as the construction information such as the improvement
length of each pile, pressure, speed, grout volume injected, etc., are
recorded, defects in pile body and inconsistent pile lengths can not be
avoided which creates potential safety hazards for highway engineer-
ing construction. This is why it is necessary to carry out pile foundation
testing.
If there are defects in the pile foundations, the quality of ground
improvement will inevitably decline. Therefore, there is a relationship
of mutual influence and restriction between pile foundation defects
and ground improvement quality. The ultimate purpose of pile foun-
dation defect detection is to evaluate whether the quality of ground
improvement has achieved good results. However, it is very challen-
ging to accurately evaluate the quality of pile construction due to the

Nature Communications | (2023)14:3448 7

Article
limitations of DCM pile material, large number of pile groups and small
pile spacing. If the single pile detection technology is adopted, and
then one pile after another is detected, not only will the detection
effect be poor, but also the detection efficiency will be low, and the
cost will be high. If the foundation strengthened by piles is regarded as
a complete set of strata, the problem of pile foundation testing can be
transformed into a problem of ground improvement quality evalua-
tion. It can significantly reduce the detection cost while effectively
improving the detection efficiency.
Piling is actually a process of transforming the original formation,
and the transformed formation medium will undergo significant
changes14,59, which also will lead to changes in the electromagnetic
properties of different underground locations. The degree of geolo-
gical transformation varies, resulting in different changes in the
underground medium and varying electromagnetic characteristics55,56.
This creates conditions for using GPR technology to detect the chan-
ges in the stratum caused pile construction and the differences in the
changes in the strata media between different underground locations.
Difference in the electromagnetic properties of underground media
can cause changes in the GPR response, specifically manifested in
changes in the waveform, amplitude, frequency, phase and related
attributes of the GPR signals40,42,49,50,54,55. However, the propagation and
response of electromagnetic wave fields under the condition of pile
groups are very complex, and it is not easy to fully reveal the propa-
gation laws of electromagnetic waves under such condition. For this
reason, we have constructed geological models of subgrade reinfor-
cement by pile groups, and qualitatively discussed the electro-
magnetic wave propagation phenomenon in this case. Compared with
the complex problem of pressure-controlled spherical cavity expan-
sion in semi-infinite soil, although our models are simplified or idea-
lized geological models, we believe that their GPR response
mechanisms are identical. That is, GPR response is based on differ-
ences in the electromagnetic properties of media. As long as there are
differences in underground media, differences in GPR response char-
acteristics will inevitably be appeared. Therefore, GPR technology is
fully applicable to the detection of geological defects caused by
pressure-controlled spherical cavity expansion in semi-infinite soil. For
the stationary oscillating waves appearing in the measured data (as
shown in Fig. 4b), we speculate that this may be due to the guided wave
phenomenon generated by electromagnetic waves within or between
piles. We believe that these waves have a positive significance for
judging the integrity of piles, and therefore can be considered as
effective wave fields.
In theory, higher frequency and smaller spacing do contribute to
improving the resolution of detection. However, if the frequency of
GPR antennas is too high, the penetration depth of electromagnetic
waves will be decreased. According to the existing technical experi-
ence of GPR40–55, the maximum effective detection depth of 100 MHz
antenna is approximately 50 m and the maximum effective detection
depth of 250 MHz antenna is approximately 10 m in the survey of
Quaternary strata. In addition, if the channel spacing is too small, it will
inevitably increase the workload, thereby increase survey cost. In our
case, we selected 100 MHz antennas for detection. The electro-
magnetic wave velocity v obtained by matching GPR section with the
drilled core was approximately 0.1 m · ns−1 and the selected sampling
rate Δt was 0.351 ns. Therefore, the detection depth h of 1024 sampling
points per channel is approximately:
h = v × 4t × n=2 = 0:1 × 0:351 × 1024=2 = 17:97m: ð1Þ
Our case shows that the data acquisition parameters, such as
100 MHz antenna, 2.0 cm point spacing, 0.351 ns sampling rate, and
1024 sampling points per channel, are fully suitable for the quality
evaluation of soft soil subgrade reinforced by DCM piles with a length
of approximately 10.0 m.
Nature Communications | (2023)14:3448
https://doi.org/10.1038/s41467-023-39236-4
The signals that cause the GPR response will be very weak when the
differences in the physical properties of underground media are quiet
weak. This makes it difficult to identify the information of formation
media differences from weak GPR signals42. It can be seen from Fig. 6e
that the traditional waveform (section) interpretation results omit to
identify the defect area D2, and it is also difficult to accurately identify
the bottom interface of the pile foundations. This is a manifestation of
inaccurate interpretation results due to insufficient utilization of
information carried by GPR data. Our research results show that the
problem of fine quality evaluation of ground improvement can be
effectively solved by integrating the GPR information of single-channel
waveform, multi-channel section and attributes. On the basis of pre-
liminary evaluation results obtained from waveform (section) inter-
pretation, comprehensive analysis and interpretation of multi-attribute
information can not only accurately identify the depth of DCM piles and
subgrade defect areas, but also further subdivide the subgrade
strengthened by pile groups into reinforced and weakly reinforced
stratum. It provides an important technical guarantee for the safe and
efficient construction of soft soil subgrade reinforcement projects.
The more data revealing a scientific or engineering problem is not
necessarily the better, but whether it is typical and representative.
Although we have only reported one GPR section, this GPR section has
already included possible problems with DCM piles reinforcement of
subgrade, such as deviation between pile length and design, defects in
pile foundations, and differences in the reinforcement quality of sub-
grade in different sections. Therefore, our sample data fully possesses
typical and representative characteristics. It should be emphasized
that our work is mainly to detect the defects of DCM piles and evaluate
the effectiveness of ground improvement based on GPR data. How-
ever, it is not rule out that there are other scientific issues in ground
improvement engineering that can be studied or explored through
GPR information. Multi-data samples may be an important way to
discover and solve unknown scientific issues.
Each technology will have more or less deficiencies (or limita-
tions), and so does our technology. Specifically, if there are strong
electromagnetic field sources (such as high-voltage lines, substations,
wireless communication base stations, etc.) or large metal objects
(such as trucks, drilling machines, etc.) in the construction site envir-
onment, GPR technology will not achieve good detection results.
Therefore, when using GPR survey, it is important to avoid these
situations as much as possible. In addition, when the electromagnetic
properties of the detection targets and the surrounding medium are
close, it is easy to cause misjudgment, that is, GPR survey may not be
applicable at all in this case. Moreover, GPR interpretation usually has
multiple solutions. The GPR response characteristics caused by dif-
ferent media are close to each other or the same media causes dif-
ferent GPR response characteristics. What kind of interpretation can
match the actual situation? The work experience of technicians will
play an important role in this case. Finally, it should be noted that our
GPR technology can effectively distinguish the electromagnetic dif-
ferences of different formation media, but it is difficult to achieve a
quantitative (or accurate) interpretation of the relationship between
the ultimate expansion pressure and the ground surface displacement
of the expansion pressure.
In summary, the effectiveness of soft soil subgrade reinforcement
depends entirely on the quality of pile construction, and defect
detection of pile foundations can be regarded as an effective mon-
itoring measure for the quality of pile construction or as an insurance
against accidents33–39. Therefore, in order to truly improve the quality
of subgrade reinforcement by DCM piles, it is necessary to strictly
follow the specifications during the pile construction process, elim-
inate all possible construction errors, and strictly control the con-
struction quality. Cutting off the possibility of pile foundation defects
from the source is the eternal goal pursued by subgrade engineering
construction.
8
Article
Methods
GPR data processing and geological information interpretation are two
key technologies that restrict the ability of GPR survey to achieve good
geological results. We have established a workflow of GPR data pro-
cessing and geological information interpretation for the defect
detection of DCM piles and quality evaluation of ground improvement
(Supplementary Fig. S3).
GPR data processing
High quality GPR data processing is the foundation and guarantee for
accurate and complete interpretation of geological information
carried by GPR data43. If the quality of GPR data processing is not
high, the geological information carried by GPR data will not be fully
extracted or there may be omissions, and the effectiveness of pile
foundation defect detection and ground improvement quality eva-
luation will be greatly reduced. According to the actual situation of
defect detection of DCM piles and ground improvement quality
evaluation as well as the characteristics of GPR data collected, in
addition to zero correction, gain recovery, inter channel and intra
channel energy compensation, it is necessary to emphasize proces-
sing such as interference noise suppression and multiple attenuation.
The purpose is to ensure high-quality GPR data processing results
without losing valid information. In addition, GPR data obtained
under undulating surface conditions require topographic correction.
Otherwise it will inevitably distort the GPR response characteristics
caused by underground geological targets, leading to distortion of
GPR interpretation results. Due to the relatively flat surface of our
test site, no topographic correction was performed during GPR data
processing in our case.
Geological information interpretation
Only by fully mining the geological information carried in GPR data can
we further accurately evaluate the quality of pile foundation and the
effectiveness of ground improvement. The changes in GPR waveform,
amplitude, frequency and phase are closely related to the physical
mechanism of electromagnetic wave propagation, geotechnical phy-
sical properties, stratigraphic structure and other factors. GPR section
interpretation is the interpretation of the geological information car-
ried by GPR section based on the changes in waveform, amplitude,
frequency and phase characteristics. It is the most basic content of
GPR data interpretation61. Firstly, based on the theoretical GPR
response characteristics and understanding obtained from the sub-
grade geological models of pile group reinforcement (Fig. 1), typical
GPR channels in the section are extracted for response characteristics
analysis (Fig. 4b), and information such as pile foundation complete
and defects are initially obtained. Then, it is extended to GPR section
interpretation (Fig. 4c) to obtain subgrade structure, defects and other
information below the entire survey line based on the conclusion of
single-channel interpretation.
The variation characteristics of waveform, amplitude, frequency
and phase only reflect one aspect of geological information carried in
GPR data. When underground reflection characteristics are obvious,
they can effectively reflect the relationship between GPR response
characteristics and stratum structure61. However, it is an indisputable
fact that the structure and physical properties of the underground are
complex and variable. When the difference in wave impedance is not
significant, the characteristics changes in waveform, amplitude, fre-
quency and phase of GPR data will be not significantly, but it does not
indicate that the stratum structure or physical properties have not
changed. Therefore, relying solely on the waveform, amplitude, fre-
quency and phase characteristics of GPR data, it is often difficult to
accurately reveal the true internal structural characteristics and phy-
sical property distribution of pile foundations and subgrade. On the
other hand, in addition to the above four basic information, GPR data
also contain rich attribute information that can be utilized. GPR
Nature Communications | (2023)14:3448
https://doi.org/10.1038/s41467-023-39236-4
attributes refer to extracting geometric, kinematic, dynamic and sta-
tistical characteristics related to electromagnetic wave response and
propagation from GPR records. GPR attribute analysis is to extract
descriptive and quantitative GPR attribute features to characterize the
structure and physical property distribution information of under-
ground targets62,63. Currently, GPR attribute information has been
applied to distinguish rock stratum interface and fault fracture areas64,
detect light non aqueous phase liquids (LNAPL) pollution65, identify
collapsed ancient caves66, high-resolution glacier imaging67 and survey
underground structure in archaeology61,68,69. These successful appli-
cation cases show that extracting GPR attribute information for ana-
lysis can fully mine the geological information carried in GPR data,
thereby enabling better interpretation of underground targets.
Therefore, we have specially developed a GPR attribute information
extraction and analysis technology for pile foundation defect detec-
tion and subgrade reinforcement quality evaluation. Here, we will
briefly introduce the four most representative GPR attribute
principles.
Arc length of time window (AL)
The AL attribute is defined as the expanded length of the waveform
curve within the analysis time window (Eq. 2). It is a joint attribute that
combines amplitude and frequency characteristics and can be used to
distinguish the phase characteristics of GPR records such as between
strong amplitude and high frequency, between strong amplitude and
low frequency, between weak amplitude and high frequency as well as
between weak amplitude and low frequency (Fig. 5a).
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 X n
AL = ½ ðAði + 1Þ  AðiÞ2 + 4t 2 , ð2Þ
n  4t i = 1
where, n represents the number of sampling points within the analysis
time window, Δt forms the sampling rate (sampling time), A(i) denotes
the amplitude value of the ith sampling point, and i is the sampling
sequence number (i = 1, 2, 3, …, n).
Δt determines the resolution of the sampled signals. A smaller
sampling time means a higher sampling resolution, but it also means
an increase in sampling data. For GPR survey at subgrade scale, we
recommend a sampling time of 0.1 to 0.4 ns. Δt was taken 0.351 ns in
our case. For n, statistical analysis is generally performed using sam-
pling points with a half wavelet length. The smaller the length of the
analysis time window, the higher the resolution of the GPR attribute
section obtained. However, the length of the statistical analysis time
window cannot be too small (such as n < 2), otherwise the statistical
analysis will lose its significance. We took a statistical analysis time
window with a length of 1.0 ns in our case, that is, took 2 data for each
time window for statistical analysis.
Product of instantaneous amplitude and cosine of instantaneous
phase (PIACIP). The PIACIP attribute strengthens the amplitudes of
wave peaks and troughs, and inverts all trough amplitudes into
apparent peak amplitudes (Fig. 5b), which is effective for analyzing
amplitude anomalies.
 
PLACIP = AInc ðt Þ  cos QInc ðt Þ , ð3Þ
where, AInc represents instantaneous amplitude, QInc denotes instan-
taneous phase, and t forms travel time, cos[] is the cosine function.
Slope of instantaneous frequency (SIF). The SIF attribute is defined
as the change rate of instantaneous frequency over time within the
analysis time window (Eq. 4). It highlights changes in local frequency,
so it can more effectively reflect differences in thin layers, and is
effective for dividing edge phases (Fig. 5c). It is also commonly used to
9
Article
indicate the rate of attenuation and absorption.
df Inc
SIF = , ð4Þ
dt
where, fInc denotes instantaneous frequency, t forms travel time, and d
represents derivation operator.
Slope of reflection strength (SRS). The SRS attribute is defined as the
change of instantaneous reflection intensity (i.e. instantaneous
amplitude) with reflection time (Eq. 5). If the reflection intensity
remains throughout the analysis time window interval, the slope value
will approach zero. If the reflection intensity gradually increases from
the top to the bottom of the time window, the slope value is positive,
on the contrary, the slope value is negative. The attribute can reflect
the vertical distribution characteristics and interbedding conditions of
the stratum (Fig. 5d).
AIns ðt Þ
SRS = , ð5Þ
dt
where, AIns represents instantaneous amplitude, t forms travel time,
and d is derivation operator.
Calculation of velocity
It is necessary to know the propagation velocity of electromagnetic
waves when conducting time-depth conversion for GPR section.
Through joint calibration of drilling results and GPR section, the buried
depth of the formation can be determined, and then the double-way
travel time of the formation interface can be picked up from the GPR
section to calculated the electromagnetic wave propagation velocity,
2h
v= , ð6Þ
t vertical bearing capacity of stiffened deep cement mixing piles. Int.
where, v represents the propagation velocity of electromagnetic
waves, h forms the depth of the formation interface, and t denotes the
double-way travel time.
Data availability
All the data supporting the findings of this study are available within
the main text and the Supplementary Information. The source data
generated in this study are provided in the Source Data file and also
available in Figshare under accession code https://doi.org/10.6084/
m9.figshare.22958399. Source data are provided with this paper.
Code availability
The source codes and respective compiled executable are provided as
Supplementary Software 1 file. The source code has also been made
available in Figshare under accession code https://doi.org/10.6084/
m9.figshare.22958399.
References
1. Weng, X. & Zhang, L. Settlement prediction of soft subgrade under
broaden embankment. J. Chang’ Univ. 31, 17–21 (2011).
2. Yang, J. & Feng, Q. A new method for measuring subgrade settle-
ment in high-speed railway by using a linear CCD. Measurement 46,
1751–1756 (2013).
3. Li, G., Nguyen, T. N. & Amenuvor, A. Settlement prediction of sur-
charge preloaded low embankment on soft ground subjected to
cyclic loading. Mar. Georesources Geotechnol. 34, 154–161 (2016).
4. Li, H., Zheng, X., Sheng, Y. & Ke, S. Differential settlements of
embankment treated by cement fly-ash gravel pile and sheet pile in
freeway extension constructions. Int. J. Geomech. 17,
40170921 (2017).
Nature Communications | (2023)14:3448
https://doi.org/10.1038/s41467-023-39236-4
5. Shan, Y. et al. Iterative method for predicting uneven settlement
caused by high-speed train loads in transition-zone subgrade. J.
Transportation Res. Board 2607, 7–14 (2017).
6. Guo, Y., Zhai, W. & Sun, Y. A mechanical model of vehicleslab track
coupled system with differential subgrade settlement. Struct. Eng.
Mech. 66, 15–25 (2018).
7. Zhou, S. & Wang, B. Review of research on high-speed railway
subgrade settlement in soft soil area. Railw. Eng. Sci. 28,
129–145 (2020).
8. Emadali, L., Motagh, M. & Haghighi, M. H. Characterizing post-
construction settlement of the Masjed-Soleyman embankment
dam, Southwest Iran, using TerraSAR-X SpotLight radar imagery.
Eng. Struct. 143, 261–273 (2017).
9. Zou, S. et al. Rheological characteristics and one-dimensional iso-
tache modelling of marine soft clays. Mar. Georesources Geo-
technol. 37, 660–670 (2019).
10. Wang, Y., Li, J. & Sun, D. Jet-grouting in ground improvement and
rotary grouting pile installation: theoretical analysis. Geomech. Eng.
21, 279–288 (2020).
11. Meng, X., Liao, H. & Zhang, J. Research on the collapsibility of loess
after water immersion. Nat. Hazards 109, 303–328 (2021).
12. Cheang, W., Bergado, D. T., Suksawat, T., Jamsawang, P. & Vootti-
pruex, P. Behavior and simulation of deep cement mixing (DCM)
and stiffened deep cement mixing (SDCM) piles under full scale
loading. Soils Found. 51, 307–320 (2011).
13. Taghavi, A., Muraleetharan, K. K. & Miller, G. A. Nonlinear seismic
behavior of pile groups in cement-improved soft clay. Soil Dyn.
Earthq. Eng. 99, 189–202 (2017).
14. Wonglert, A., Jongpradist, P., Jamsawang, P. & Larsson, S.
Bearing capacity and failure behaviors of floating stiffened deep
cement mixing columns under axial load. Soils Found. 58, 446–461
(2018).
15. Zhou, M., Li, Z., Han, Y., Ni, P. & Wang, Y. Experimental study on the
J. Geomech. 22, 04022043 (2022).
16. Zhu, S., Chen, C., Cai, H. & Mao, F. Analytical modeling for the load-
transfer behavior of stiffened deep cement mixing (SDCM) pile with
rigid cap in layer soils. Comput. Geotech. 144, 104618 (2022).
17. Tan, T. S., Goh, T. L. & Yong, K. Y. Properties of Singapore marine
clays improved by cement mixing. Geotech. Test. J. 22,
422–433 (2002).
18. Ignat, R., Baker, S., Larsson, S. & Liedberg, S. Two- and three-
dimensional analyses of excavation support with rows of dry deep
mixing columns. Comput. Geotech. 66, 16–30 (2015).
19. Wu, Y., Diao, H., Jian, L. & Zeng, C. Field studies of a technique to
mitigate ground settlement of operating highways. J. Zhejiang Univ.
-Sci. A: Appl. Phys. Eng. 17, 565–576 (2016).
20. Sun, Y. et al. Model test of the combined subgrade treatment by
hydraulic sand fills and soil-cement mixing piles. Bull. Eng. Geol.
Environ. 79, 2907–2918 (2020).
21. Moghadam, M. S., Zad, A., Yazdi, M. & Moghadam, M. J. Perfor-
mance of T-shaped and conventional cement-soil deep mixing piles
to stabilize soft base of high-speed trains. Int. J. Geotech. Eng. 16,
208–222 (2022).
22. Phutthananon, C. et al. Reliability-based settlement analysis of
embankments over soft soils reinforced with T-shaped deep
cement mixing piles. Front. Struct. Civ. Eng. 16, 683–656 (2022).
23. Shu, B. et al. Case study of solid waste based soft soil solidifying
materials applied in deep mixing pile. Buildings 12, 1193 (2022).
24. Jia, J., Zhao, Y., Jia, J., Xin, C. & Zhang, B. Research on the reinfor-
cement effect of cement-soil mixing pile on collapsible loess
foundation. J. Railw. Eng. Soc. 39, 18–24 (2022).
25. Wonglert, A. & Jongpradist, P. Impact of reinforced core on per-
formance and failure behavior of stiffened deep cement mixing
piles. Computers Geotech. 69, 93–104 (2015).
10
Article
26. Fleming, B. J., Sritharan, S., Miller, G. A. & Muraleetharan, K. K. Full-
Scale seismic testing of piles in improved and unimproved soft clay.
Earthq. Spectra. Prof. J. Earthq. Eng. Res. Inst. 32, 239–265 (2016).
27. Briaud, J. L., Ballouz, M. & Nasr, G. J. Static capacity prediction by
dynamic methods for three bored piles. J. Geotech. Geoenviron.
Eng. 126, 1078–1084 (2001).
28. Sakr, M. Comparison between high strain dynamic and static load
tests of helical piles in cohesive soils. Soil Dyn. Earthq. Eng. 54,
20–30 (2013).
29. Mostafa, H. H. & Elhakim, A. F. Static versus dynamic load tests on
piles socketed in Dubai limestone. Innovative Infrastruct. Solut. 7,
298 (2022).
30. Hou, Y., Mi, X. & Hong, X. The origin identification on defect cement
mixing pile in the core sampling. Int. Conf. Electr. Technol. Civ Eng,
Lushan, China (2011).
31. Gong, D. et al. Coring of antarctic subglacial sediments. J. Mar. Sci.
Eng. 7, 194 (2019).
32. Wrana, B. Pile load capacity: Calculation method. Stud. Geo-
technica et. Mechanica 37, 83–93 (2015).
33. Alwalan, M. F. & El Naggar, M. H. Analytical models of impact
force-time response generated from high strain dynamic load test
on driven and helical piles. Computers Geotech. 128, 103834
(2020).
34. Tu, Y., El Naggar, M. H. & Wang, K. Nonlinear fictitious-soil pile
model for pile high-strain dynamic analysis. Soil Dyn. Earthq. Eng.
162, 107464 (2022).
35. Dhutti, A. et al. Numerical modeling of ultrasonic guided wave
propagation and defect detection in offshore steel sheet piles.
Appl. Sci. 11, 4076 (2021).
36. Guan, W. et al. Influence of the three-dimensional effect of pile-soil
system on the vertical dynamic response of large-diameter piles in
low-strain integrity testing. Energies 15, 9548 (2022).
37. Liao, S. T. & Roesset, J. M. Identification of defects in piles through
dynamic testing. Int. J. Numer. Anal. Methods Geomech. 21,
277–291 (1997).
38. Chow, Y. K., Phoon, K. K., Chow, W. F. & Wong, K. Y. Low strain
integrity testing of piles: three-dimensional effects. J. Geotech.
Geoenviron. Eng. 129, 1057–1062 (2003).
39. Wu, W. & Zhang, Y. A review of pile foundations in viscoelastic
medium: dynamic analysis and wave propagation modeling. Ener-
gies 15, 9432 (2022).
40. Davis, J. L. & Annan, A. P. Ground-penetrating radar for high-
resolution mapping of soil and rock stratigraphy. Geophys. Pro-
spect. 37, 531–551 (1989).
41. Bradford, J. H., Privette, J., Wilkins, D. & Ford, R. Reverse-time
migration from rugged topography to image ground-penetrating
radar data in complex environments. Engineering 4,
661–666 (2018).
42. Shen, H. et al. Enhancement of GPR diffracted waves processing: a
case study from Taiyuan, China. J. Environ. Eng. Geophys. 24,
237–247 (2019).
43. Lombardi, F., Podd, F. & Solla, M. From its core to the niche: insights
from GPR applications. Remote Sens. 14, 3033 (2022).
44. Aranha, P. R. A., Augustin, C. H. R. R. & Sobreira, F. G. The use of GPR
for characterizing underground weathered profiles in the sub-
humid tropics. J. Appl. Geophys. 49, 195–210 (2002).
45. Böniger, U. & Tronicke, J. Integrated data analysis at an archae-
ological site: A case study using 3D GPR, magnetic, and high-
resolution topographic data. Geophysics 75, B169–B176 (2010).
46. Hebsur, A. V., Muniappan, N., Rao, E. P. & Venkatachalam, G.
Application of ground penetrating radar for locating buried impe-
diments to geotechnical exploration and piling. Int. J. Geotech. Eng.
7, 374–387 (2013).
47. Conyers, L. B. Analysis and interpretation of GPR datasets for inte-
grated archaeological mapping. Surf. Geophys. 13, 654–651 (2015).
Nature Communications | (2023)14:3448
https://doi.org/10.1038/s41467-023-39236-4
48. Schneiderwind, S., Mason, J., Wiatr, T., Papanikolaou, I. & Reicherter,
K. 3-D visualisation of palaeoseismic trench stratigraphy and trench
logging using terrestrial remote sensing and GPR: a multiparametric
interpretation. Solid Earth 7, 323–340 (2016).
49. Torbaghan, M. E. et al. Automated detection of cracks in roads using
ground penetrating radar. J. Appl. Geophys. 179, 104118 (2020).
50. Pilecki, Z. et al. Identification of buried historical mineshaft using
ground- penetrating radar. Eng. Geol. 294, 106400 (2021).
51. Yamaguchi, T. & Mizutani, T. Detection and localization of manhole
and joint covers in radar images by support vector machine and
Hough transform. Autom. Constr. 126, 103651 (2021).
52. Bano, M., Tsend-Ayush, N., Schlupp, A. & Munkhuu, U. Ground-
penetrating radar imaging of near-surface deformation along the
Songino active fault in the Vicinity of Ulaanbaatar, Mongolia. Appl.
Sci. 11, 82429 (2021).
53. Li, C. et al. Layered subsurface in Utopia Basin of Mars revealed by
Zhurong rover radar. Nature 610, 308–321 (2022).
54. Noshahri, H., Van der Meijde, M. & Scholtenhuis, L. O. GPR surveys
in enclosed underground sewer pipe space. Tunn. Undergr. Space
Technol. Inc. Trenchless Technol. Res. 129, 104689 (2022).
55. Tesic, K., Baricevic, A., Serdar, M. & Gucunski, N. Characterization of
ground penetrating radar signal during simulated corrosion of
concrete reinforcement. Autom. Constr. 143, 104548 (2022).
56. Dong, X., Song, Z., Zhang, S., Yang, B. & Wu, S. Application study of
electrical resistivity characteristics of core samples from soil-
cement mixing piles. China Civ. Eng. J. 49, 88–94 (2016).
57. Si, F. et al. Experimental study on diameter detection of high
pressure jet grouting pile based on electromagnetic induction
effect. Chin. J. Rock. Mech. Eng. 40, 3434–3442 (2021).
58. Yang, Y., Liu, Y. & Ren, X. Experimental research on the factors
influencing strength of cement mixed saturated loess. J. Railw. Eng.
Soc. 33, 21–25 (2016).
59. Zhou, Z., Wang, K., Feng, H., Tian, Y. & Zhu, S. Centrifugal model test
of post-grouting pile group in loess area. Soil Dyn. Earthq. Eng. 151,
106985 (2021).
60. Zhou, M., Xing, L., Chen, Y. & Qiao, H. Settlement characteristics of
split grouting pile composite foundation in loess areas. Adv. Civ.
Eng. 2022, 8627124 (2022).
61. Manataki, M., Vafidis, A. & Sarris, A. GPR data interpretation
approaches in archaeological prospection. Appl. Sci. 11, 7531 (2021).
62. Tronicke, J. & Böniger, U. GPR attribute analysis: there is more than
amplitudes. First Break 31, 103–108 (2013).
63. Zhao, W., Forte, E., Pipan, M. & Tian, G. Ground Penetrating Radar
(GPR) attribute analysis for archaeological prospection. J. Appl.
Geophys. 97, 107–117 (2013).
64. Jorge, L. P., Sauck, W. A. & Junior, A. O. S. GPR for mapping fractures
and as a guide for the extraction of ornamental granite from a
quarry: a case study from southern Brazil. J. Appl. Geophys. 58,
177–187 (2006).
65. Bradford, J. H. & Wu, Y. Instantaneous spectral analysis: time-
frequency mapping via wavelet matching with application to
contaminated-site characterization by 3D GPR. Lead. Edge 26,
1018–1023 (2007).
66. dos Reis Júnior, J. A., de Castro, D. L., de Jesus, T. D. S. & Lima-Filho,
F. P. Characterization of collapsed paleocave systems using GPR
attributes. J. Appl. Geophys. 103, 43–56 (2014).
67. Zhao, W., Forte, E., Colucci, R. R. & Pipan, M. High-resolution glacier
imaging and characterization by means of GPR attribute analysis.
Geophys. J. Int. 206, 1366–1374 (2016).
68. Khwanmuang, W. & Udphuay, S. Ground-penetrating radar attri-
bute analysis for visualization of subsurface archaeological struc-
tures. Lead. Edge 31, 946–949 (2011).
69. Zhao, W., Forte, E. & Pipan, M. Texture attribute analysis of GPR data
for archaeological prospection. Pure Appl. Geophys. 173,
2737–2751 (2016).
11
Article
Acknowledgements
The authors thank Professor Qingchun Li at Chang′an University, Professor
Chengqian Tan and Professor Zhiping Zhang at Xi′an Shiyou University for
valuable suggestions. The authors also thank the four anonymous
reviewers for constructive comments and recommendations. H.S. thanks
the Natural Science Basic Research Program of Shaanxi Province of China
for Project 2017JZ007 and the Key Research and Development Project of
Shaanxi Province of China for Project 2022GY-148.
Author contributions
H.S., X.L., R.D. and Y.Z. devised the initial concept for the work. H.S., X.L.
and R.D. designed the experiments. H.S., R.D. and Y.Y. developed the
data processing software. R.D., X.L., Z.X., J.L., B.L., Ho.C. and J.G. col-
lected the data. H.S., R.D., Y.Z., J.Z., Ha.C. and Y.Y. processed and ana-
lyzed the data. H.S., G.L. and Ha.C. drew the figures and table. H.S., X.L.
and R.D. co-wrote the manuscript. H.S., X.L., R.D., Y.Z., J.Z., Ha.C., Z.X.,
C.Y., J.L., C.J. and Y.Y. discussed the results and commented on the
manuscript at all stages.
Competing interests
The authors declare no competing interests.
Additional information
Supplementary information The online version contains
supplementary material available at
https://doi.org/10.1038/s41467-023-39236-4.
Nature Communications | (2023)14:3448
https://doi.org/10.1038/s41467-023-39236-4
Correspondence and requests for materials should be addressed to
Hongyan Shen, Ruifeng Duan or Yueying Yan.
Peer review information Nature Communications thanks Vivi Anggraini,
Takahiro Yamaguchi and the other, anonymous, reviewer(s) for their
contribution to the peer review of this work. A peer review file is avail-
able.
Reprints and permissions information is available at
http://www.nature.com/reprints
Publisher’s note Springer Nature remains neutral with regard to jur-
isdictional claims in published maps and institutional affiliations.
Open Access This article is licensed under a Creative Commons
Attribution 4.0 International License, which permits use, sharing,
adaptation, distribution and reproduction in any medium or format, as
long as you give appropriate credit to the original author(s) and the
source, provide a link to the Creative Commons license, and indicate if
changes were made. The images or other third party material in this
article are included in the article’s Creative Commons license, unless
indicated otherwise in a credit line to the material. If material is not
included in the article’s Creative Commons license and your intended
use is not permitted by statutory regulation or exceeds the permitted
use, you will need to obtain permission directly from the copyright
holder. To view a copy of this license, visit http://creativecommons.org/
licenses/by/4.0/.
© The Author(s) 2023
12